MODELLING THE NUMBER OF HIV CASES IN INDONESIA USING NEGATIVE BINOMIAL REGRESSION BASED ON LEAST SQUARE SPLINE ESTIMATOR

HIV is a quite dangerous infectious disease. The HIV transmission can be controlled through several factors including the number of prisoners, early-age marriages, and contraceptive users. In order to control these factors, proper judgment must be used. To analyze the association between HIV and these factors, an appropriate statistical model approach, such as, a least square spline nonparametric negative binomial regression model, is proposed in this research. The results of the analysis show that the proposed model approach is better than the classical model approach, namely, parametric negative binomial regression model. This is supported by a deviance value of the proposed model of 15.6552 which is less than a deviance value of the parametric negative binomial regression model that is 32.2954. Also, the number of HIV cases will decrease when the number of prisoners is less than 768. In addition, the number of the HIV cases will increase only if the percentage of early-ager marriages is between 11.155% and 3.2475%, and apart from this range the number of HIV cases will decrease. Additionally, the number of HIV cases will increase if contraceptive users are between 42.085% and 48.610%, apart from this range the number of HIV cases will decrease.